KIH #3: Can digital volume controls keep it honest?

Digital volume control. It’s becoming all the rage. Digital amplifier used to be a misnomer as though that’s what class D meant. In truth such amps were amps with analog inputs, zero D/A conversion and switching analog output stages. SMPS or switch-mode power supplies were entirely optional. Nothing digital about any of it. Sticking a volume control to such an amp turned it into an integrated with, yes you guessed right, analog volume control.

But when the whole computer audio thing took off, DACs regained traction. Soon those began to offer volume control often via on-chip DSP. And why not? Unless you had analog sources, preamps were finally redundant and passé. Or were they? Next came digital power processing chips which could convert incoming PCM data to PWM to drive switching output devices directly. Now we had true digital amplifiers with necessarily digital volume. Devialet, NAD, NuForce and Wadia showed how that’s done.

In all the excitement over integration, simplification, reduced box and cable clutter and the very real promise of better performance from it all, one thing gets routinely overlooked. Common digital volume controls become lossy. The worst examples arise when hi-gain converters which put out a full-scale 4V or more are mated to amps with high gain of their own (say 30dB) and the lot is played into normally efficient speakers in standard rooms at regular to low volumes.

My room is 5.5m x 12m WxD. My various speakers live between roughly 89 – 93dB/1W/1m. With an amp like Goldmund’s affordable Job 225, AURALiC’s brilliant Vega processor will easily get pushed down into 20 on its dial of a max 100. Because its volume is Sabre’s on-chip digital solution, such deep attenuation is plainly audible. The sound gets pale, thin and flat by contrast to its full-scale splendour. Should you ask what my DAC sounds like, I’d have to say it depends on where its volume sits. That’s why I still use my very expensive Nagra Jazz valve preamp to do volume. The Vega never leaves 100.

One obvious solution around this is a low-gain amp. Finding a transistor version with less than 26dB of gain is simply not easy. Another solution would be the seriously inefficient speaker, perhaps a small very bass-capable monitor like a Mark & Daniel or Prime Loudspeakers. I personally think that particular approach is rather less elegant and a bit brutish but it would certainly work. You’d invoke far less digital cut to remain in a more non-critical range.

Another solution is something Daniel Weiss offered already on my old pro-model DAC2. An analog trim pot could set its full-scale output down from 2V to 1V and 0.5V (I don’t remember the exact values but you get the idea). This forced the digital volume to operate in its uppermost quadrant. Simple but very effective.

Why don’t AURALiC, Resonessence & Co. as leaders of truly attractive converters do this? As is their decks cannot satisfactorily replace analog preamps unless some very specific conditions were met first. A multi-stage analog gain trim coupled to their present 0.5/1dB-step digital controls would immediately relax or obliterate those conditions.

Here Antelope Audio’s Zodiac Gold and Platinum models go the whole hog . Their volume is purely analog with relay-switched resistor-ladder attenuators. This even enables addition of true analog inputs without redundant A/D/A conversion. That’s one proper way to eliminate the preamp.

How about true digital amplifiers? By definition they must employ digital volume. To eliminate or minimize audible losses regardless of setting, they limit their circuit’s overall voltage gain to—with most speakers, rooms and listening habits (here some conditions enter again)–remain bona fide high-resolution machines.

It seems that a future trend in this sector will be 64-bit math. Software players like PureMusic have offered it for years. Exploiting your computer’s processing power which for audio purposes is extreme, upsampling and/or digital volume in such software gives better results than dumping the same tasks onto 24-bit or 32-bit audio chips.

One company that recently moved 64-bit math out of the computer into their own hardware is Bel Canto. Their new Black http://www.belcantoblack.com/ range’s “64-bit digital signal processor core provides complete control over the signal processing functions, allowing Black optimal control of volume/balance, input selection, custom filters and the asynchronous interfaces all without computational compromise…”

An additional step which BCD took as you read the small print is running a custom low-gain version of Bruno Putzeys’ Ncore 1200 power modules. This assures that even their hi-zoot digital volume doesn’t get stressed into super deep attenuation mode. Antelope Audio’s proprietary jitter-reclocking algorithms and upsamplers too exploit 64-bit math in custom-coded field-programmable gate arrays (FPGA) to transcend the current limits of 24-bit on-chip solutions.

Keeping it honest on the subject of digital volume, unless current 24-bit solutions are employed within predefined parameters of overall gain, speaker sensitivity and playback levels, they’re lossy and inferior to quality analog controls. That said it’s just as true that such losses might be offset by gains made elsewhere. Shorter signal paths, fewer gain stages and less cables from running DAC to amp direct could have the bigger benefits. It simply wouldn’t mean that digital volume isn’t lossy. It only means that its losses are less severe than the alternative.

How in a super-resolving context even 0.5dB of digital attenuation can be audible was brought home at the recent HighEnd Swiss show in Zürich. Angel Despotov of luxury amplifier firm Analog Domain showed with Göbel loudspeakers. Despite the usual small hotel room their sound was of extremely high resolution. Running off a top-pedigree dCS source, Angel demonstrated how the slightest engagement of its volume control—which we can assume was executed to a very high level—was immediately obvious as a filtering effect. It sounded like an opacity command by contrast to his own analog controller.

I very much doubt you’d have heard this in a less resolving system. But it did go to show that at this juncture and level, work remains. Despite the convenience and elegance of numerical displays and attenuation progressions in 100 x 0.5dB steps or such, most current digital volume controls aren’t as perfect as many makers claim. That doesn’t mean you ought to shun them. Simply verify that in your particular application they’re actually preferable to a good analog equivalent, even a passive one. If you must listen loud before things sound good, you probably have an issue. One possible culprit is just what we talked of today.

On the subject of digital vs. analog volume and source direct vs. a preamp, I’ll close out with this reader email exchange: “Hi Srajan, since our last email I’ve acquired a pair of used active ATC SCM100A and am now trying to find a ‘budget’ preamp to go with them at least for now. From amongst these preamps which one would you think is a reasonable mate for the 100A (input impedance >10k ohms)? They are comparably priced and with enough I/O for my use: Wyred4Sound mPre, Khozmo balanced passive, used PS Audio PCA2 with separate power supply. I am driving the 100A direct with my Oppo 105 and in general am happy with the setup but it doesn’t give me usable volume range and the needed connectivity so a simple good quality solution will do. I listen at low volume most of the time like you do, I already have the Oppo which does a decent job at DA conversion so I would guess a nice volume is more beneficial but I’m no expert on this. Your comments would be most welcome and needed.” – Raymond

I’d told Raymond to “go with the Wyred because, below unity gain, it works as a passive (like the Khozmo) but unlike the Khozmo, it’s still an impedance-compensated actively buffered ‘passive’ which gives superior performance particularly at lower volumes.”

In due time Raymond reported that “in the end I got myself a Wyred STP-SE and am happily enjoying my music. With a preamp there is more body and color versus the very nice and clear but rather plain Oppo/source direct driven setup.”

Which I wrapped with“good man. It’s common thinking that a preamp these days is redundant in many scenarios (it’s a very reasonable belief to entertain after all) and just as common to hear that most people who actually try it still prefer a preamp over source-direct and for exactly the reasons you just gave.” This doesn’t mean you’d agree. But it does mean you should definitely try this rather than rely on theory alone. Bookish audio knowledge tends to not go that far…

[You can read more on whether or not a pre-amplifier bests a direct digital setup here and here.]